v1.13.0

Astrological authority and Muhurtam coverage release. The daily Panchangam,
Daily Horoscope and Muhurtam surfaces now disclose criterion-level provenance
instead of treating computed output as uniform textual authority.

Release highlights

• Muhurtam coverage: 34 of 35 distinct purpose-specific profiles now have
  verified rule-level source claims; Anything auspicious remains the sole
  explicit heuristic. House warming is the narrowly scoped, source-verified
  Gruhapravesha election, with completed-house purchase and renovation kept
  separate.
• Corrected elections: wedding, Gruhapravesha, court filing, engagement,
  Homahuti, deferred Pretakriya, Seemantha, borrowing, lending, inventory
  purchase, service entry, capital deployment, Shantika/Paushtika and
  Dharma-kriya rules now follow their inspected passages and disclose manual
  chart conditions.
• Panchangam and horoscope corrections: Bhadra Mukha/Puchha follows the
  Tithi-specific Muhurta Chintamani Yamas; Rahu and Ketu transit houses follow
  Phaladeepika rather than inheriting Shani's set.
• MCP trust contract: find_muhurta returns the selected activity's
  provenance, automated constraints and manual prerequisites. Panchangam
  responses disclose authority state by output category.
• Clearer recommendations: browser results remove irrelevant weekday
  guidance, separate computed strengths from chart validation and practical
  information, and keep specialist Dharma-kriya/Pretakriya activities out of
  the general public selector while retaining them in Python and MCP.
• Engineering: generated Python-to-browser activity contracts, provenance
  validation, documentation-drift checks and tools/verify_project.py provide
  one fail-fast verification path across Python, frontend tests, type checking
  and production build.

Added

• Precisely scoped vehicle-acquisition profile. Raman Chapter IV, printed
  page 10 (PDF page 14), now supplies a +1 preference for five movable
  Nakshatras without treating every other star as rejected.
  Other vehicle ranking preferences remain explicitly uncited project rules;
  the citation is not presented as a complete vehicle election. Coverage is
  now 10/30.
• Source-backed Upanayana profile. Raman Chapter VIII, printed pages
  23–24 (PDF pages 27–28), now supplies lunar months, Uttarayana solar Rasis,
  exact Paksha-qualified Tithis, weekdays, Nakshatras, Lagnas and a before-noon
  gate. Age, conditional Budha combustion and chart checks remain explicit.
  Verified activity coverage reaches 9/30.
• Source-backed Vidyarambha profile. Raman Chapter VIII, printed page 23
  (PDF page 27), now supplies exact weekday, Nakshatra and movable-or-common
  Lagna gates; age, forenoon/noon preference and chart checks remain explicit.
  Verified activity coverage reaches 8/30.
• Source-backed Mundana / Chaula profile. Raman Chapter VIII, printed
  pages 22–23 (PDF pages 26–27), now supplies Shukla-Paksha, combustion,
  Tithi, Nakshatra, weekday, conservative Lagna and forenoon constraints.
  Python and browser share the solar-noon boundary. Coverage reaches 7/30.
• Source-backed Karnavedha profile. Raman Chapter VIII, printed page 22
  (PDF page 26), now supplies the never-at-night rule, weekdays and rejected
  rising Rasis; its Tithi reference resolves to Chapter V. Age, transition and
  election-house checks remain explicit. Coverage rises from 5/30 to 6/30.
• Source-backed Annaprasana profile. Raman Chapter VIII, printed pages
  21–22 (PDF pages 25–26), now supplies the admitted Nakshatras, weekdays and
  rising Rasis; its general-Tithi reference is resolved to Chapter V's exact
  Panchanga-Suddhi list. Child-age and election-chart checks remain explicit.
  Verified activity coverage rises from 4/30 to 5/30.
• Source-backed Namakarana profile. Raman Chapter VIII, printed page 21
  (PDF page 25), now supplies the ceremony's admitted Nakshatras and weekdays,
  rejected Tithi numbers, fixed-Lagna preference, and disclosed birth-day,
  election-chart and name-selection checks. Verified activity coverage rises
  from 3/30 to 4/30.
• One-command project verification. python tools/verify_project.py now
  runs provenance-link validation, generated activity-data drift checks, the
  complete Python suite, frontend tests, typecheck, and production build in a
  fixed fail-fast order.
• Machine-checked activity provenance. Verified Muhurtam profiles now
  declare their stable ledger claim beside the Python rules; MCP and the
  browser consume that shared field. A CI-ready audit rejects missing,
  non-Muhurtam, non-verified, or duplicate links and reports the honest
  coverage baseline: 3 of 30 activity profiles have exact rule locators.
• Source-backed land-purchase profile. The former generic Property / Land
  option is now accurately scoped to buying land for building, following
  Raman Chapter XII, printed page 53 (PDF page 57): fourteen Nakshatras, four
  weekdays, hard Rikta-Tithi rejection, fixed-Lagna preference, and disclosed
  manual election-chart checks. Buying a completed house is explicitly outside
  this profile rather than silently conflated with land purchase.
• Source-backed well-digging activity. Raman, Muhurtha, Chapter XII,
  printed page 51 (PDF page 55), now supplies the eight admitted Nakshatras,
  three admitted rising Rasis, the Surya-in-Lagna hard-rock caution, and
  explicit manual Shukra/Chandra chart checks. The activity is now selectable
  on the website and exposed through the same generated contract as Python.
• Source-backed Bhumi Puja / house-foundation profile. B. V. Raman,
  Muhurtha, Chapter XII, printed page 50 (PDF page 54), now governs the
  activity's admitted lunar months, solar-Rasi classes, exact Tithi numbers,
  Nakshatras, weekdays and fixed Lagna. Full election-chart and site-placement
  conditions remain explicit manual checks.
• find_muhurta now returns an additive activity_profile with the automated
  constraints, manual review requirements, and stable provenance claim ID.
  The website consumes the same generated Python activity contract.

Changed

• Bhumi Puja / foundation candidates use the cited activity profile instead
  of the former generic Bhadra-tithi/fixed-Lagna preference pair. Shared
  samskara and ranking rules remain independently disclosed.

Documentation

• Corrected the Muhurtam architecture guide from the retired
  Choghadiya-derived slot model to the named, indivisible 30-Muhurta pipeline;
  narrowed Drik Panchang's role to external Drik-output comparison rather than
  universal textual authority.
• Added a criterion-level foundation-laying crosswalk and verified
  muhurta.bhumi_puja.foundation provenance claim.
• CHANGELOG reconciled against the full git history. The operational-
  maturity round that actually shipped inside the v1.9.0 tag is now
  folded into the [1.9.0] entry — it had been stranded under a misplaced
  [Unreleased] heading below newer releases. Every release back to the
  project's first tag (v1.0.5) is now documented here, with
  comparison links.

Housekeeping

• The 15 automated-assistant cleanup commits from the 1.9.0 round were
  re-authored to the maintainer identity, so the assistant bot no longer
  appears in the contributor list. Author/committer metadata only — every
  tree SHA is unchanged, so released artifacts are byte-identical.

v1.12.0

Muhurta finder rebuilt on the named 30-muhurta grid. 1082 tests passing.

Changed

• Muhurta slots are now the named muhurtas, not choghadiya-derived
  windows. day_slots/night_slots iterate the 30 muhurtas of the
  ahoratra — 15 daytime (sunrise→sunset ÷15) + 15 night (sunset→next
  sunrise ÷15), each an indivisible ~48-min unit carrying its name,
  presiding deity, and intrinsic nature. New module
  telugu_panchangam/muhurtas.py; owner-verified reference table at
  docs/reference/07-muhurta-table.md.
• Additive, disclosed scoring. A slot's score layers: intrinsic
  muhurta nature (Abhijit/Brahma +2, other auspicious +1, inauspicious −2)
  • dominant choghadiya (a scoring attribute — straddle disclosed, not a
    gate) + weekday Durmuhurtam / hard-window exclusion + the existing
    tarabalam, chandrabalam, lagna, tithi-class and special-yoga factors.
    Abhijit and Brahma are now scored as the 8th-day / 14th-night muhurta
    natures (the old "overlaps Abhijit/Brahma" bonuses are retired; Abhijit
    is correctly absent on Wednesday). This supersedes the interim
    block-iteration fix (PR #135) and fully resolves the 2026-07-21
    choghadiya mislabel (a Kaal slot scored as Amrit).
• Each slot now leads with its muhurta identity, e.g.
  Vidhi (Abhijit) muhurta · Brahma · auspicious (+2). The website finder
  (src/panels/tarabalam.ts) mirrors the Python source of truth.
• Card wording cleaned up. User-facing muhurta reasons, skip
  diagnostics and doctrinal notes now use middots and plain punctuation
  instead of em-dashes; time and date ranges read "to". The redundant
  "clear of all inauspicious windows" line is removed: it was a tautology
  (a slot is only kept when it clears the hard windows) that also
  contradicted any muhurta of inauspicious nature. An adversarial sweep
  over 160 date/city/activity/profile combinations confirms zero dashes,
  zero contradictions, and each score equal to the sum of its (+n)/(−n)
  bonuses.
• Day context on every slot card (website). Each card carries the
  day's sunrise anga (Tithi, Nakshatra, Nitya Yoga) plus a timings row.
  Sunrise, Abhijit and Rahu Kalam show by default; the complete Panchangam
  auspicious and avoid window lists sit behind an "all timings" expand, so
  a slot's day quality is legible without switching to the Panchangam tab.
  Display only.

Affects tool_find_muhurta (MCP) and the website muhurta finder.

v1.11.0

The theme: muhurta scorer enhancements — cognitive split of the
scorer into three files, Siddha Yoga detection, Rikta dosha neutralization,
and activity-aware tithi-avoid scoring. No frozen-core or ICS-feed changes.
1055 tests passing.

Added

• Siddha Yoga detection (special_yogas.py) — all five classical
  Tithi×Vara combinations from Muhurta Chintamani now detected:
  Nanda+Friday, Bhadra+Wednesday, Jaya+Tuesday, Rikta+Saturday,
  Purna+Thursday. Scores +1 per slot. The earlier implementation was
  missing Rikta+Saturday; the incorrect comment ("Rikta has no auspicious
  Vara partner") is removed.
• Rikta dosha neutralization — classical conditions reduce the Rikta
  penalty inside score_tithi_class(): Pushya nakshatra fully cancels
  (−2 → 0); Sarvartha or Amrita Siddhi Yoga partially offsets (−2 → −1).
  Both conditions surface as explicit reason notes.
• Activity-aware tithi-avoid scoring — score_tithi_class() extended
  with avoid_tithi_class: activities declare tithi families whose energy
  is classically mismatched, scoring −1 instead of the previous flat zero.
  Jaya (combat/victory energy) avoided by: ceremony, wedding, engagement,
  naming, annaprasana, karnavedha, mundana, upanayana, vidyarambha,
  gruhapravesha, yajna. Purna (completion energy) avoided by: court,
  litigation.

Changed

• Muhurta scorer split — personal/muhurta.py (~1,200 lines) split
  into three focused files:
  • personal/activity_rules.py — declarative per-activity config
    (ACTIVITY_RULES dict, 28 activities); pure data, no logic.
  • personal/slot_scorers.py — atomic scoring functions + _DayContext
    dataclass; score_tithi_class, score_tara, score_chandra,
    score_lagna, score_special_yogas, score_nitya_yoga, etc.
  • personal/muhurta.py — orchestrator (~530 lines); public API
    (day_slots, diagnose_day, assign_tiers) unchanged.
• _DayContext dataclass — bundles 21 day-constant parameters so
  _evaluate_slot() takes (s, e, block, base, facts, ctx) instead of
  20+ positional arguments.
• _day_skip_reason() helper — day-skip logic extracted and shared
  between day_slots() and diagnose_day(), eliminating duplication.
• MCP disclaimer updated — _MUHURTA_DISCLAIMER in tools.py now
  reflects Siddha Yoga +1, symmetric tithi scoring (+1/−1/−2), and
  Pushya/Siddhi neutralization.

v1.10.4

The theme: ayanamsa full support across gochara, ingress, and phalalu tools.
get_graha_positions, get_gochara, get_rasi_phalalu, and
get_rashi_ingresses previously accepted an ayanamsa parameter but
always computed in Lahiri. All four now apply it end-to-end.
+9 tests (1054 → 1063 passed).

Changed

• get_graha_positions — ayanamsa parameter now applied to all nine
  graha longitude and ingress calculations (previously Lahiri-only with a
  warning note in the response). Stale ayanamsa_note field removed.
• get_gochara — ayanamsa parameter added and applied; graha
  positions are now computed under the requested ayanamsa.
• get_rasi_phalalu — ayanamsa parameter added and applied; reading
  is rendered from positions under the requested ayanamsa.
• get_rashi_ingresses — ayanamsa parameter added and applied;
  sign-boundary crossings are computed under the requested ayanamsa.
• gochara/positions.py refactored: graha_positions() owns the Swiss
  Ephemeris sid mode (set once, restored in finally); internal helpers
  are now mode-agnostic, avoiding redundant set_sid_mode calls on every
  loop iteration of the ingress search.
• ingress.py refactored: same own-the-mode pattern in rashi_ingresses().

Documentation

• README and PyPI page updated: tool signatures show ayanamsa="lahiri" for
  all seven ayanamsa-aware tools; new Ayanamsa section lists the four
  supported keys and clarifies SS/Vakya accept-but-ignore behaviour.

(PR #105)

v1.10.3

The theme: astronomical timing computations — four new MCP tools.
All-planet combustion/visibility calendar, planetary war detection,
rashi ingress + eclipse calendar, and five-limb Panchanga Shuddhi
assessment. Every tool is a new module consuming engine outputs;
the frozen core and ICS feeds are untouched. +84 tests
(991 → 1054 passed, 63 previously counted across PRs 101–103).

Added

• get_combustion_calendar MCP tool — heliacal Asta (setting) and
  Udaya (rising) periods for the five classical planets (Mercury, Venus,
  Mars, Jupiter, Saturn) via swe.heliacal_ut(), matching the
  Drik Panchang Asta/Udaya calendar. Per-city altitude used for
  sky-visibility accuracy. New module telugu_panchangam/maudhya_calendar.py.
  (PR #101).
• get_graha_yuddha MCP tool — Graha Yuddha (planetary war) periods:
  two tara grahas within 1° ecliptic longitude, winner by higher ecliptic
  latitude, entry/exit via binary search + ternary-search minimum.
  New module telugu_panchangam/graha_yuddha.py.
  (PR #102).
• get_rashi_ingresses MCP tool — all rashi (sign) ingress events
  for Sun, Mercury, Venus, Mars, Jupiter, Saturn, Rahu, Ketu; Lahiri
  sidereal; retrograde ingresses included; adaptive-step scan + 44-iter
  bisection. New module telugu_panchangam/ingress.py.
  (PR #103).
• get_eclipse_calendar MCP tool — solar and lunar eclipses in a
  date range with per-city visibility and Sutak timing (12 h Solar,
  9 h Lunar). Wraps the existing eclipse engine.
  (PR #103).
• get_panchanga_shuddhi MCP tool — five-limb purity assessment
  (Tithi, Vaara, Nakshatra, Yoga, Karana) with quality
  (shuddha / ashuddha / mixed) and one-line reason per limb; overall
  verdict from Sarva Ashuddha (0) to Sarva Shuddha (5). New module
  telugu_panchangam/panchanga_shuddhi.py.
  (PR #104).
• Location.alt — altitude (metres) added to the Location dataclass;
  all 22 bundled cities updated with real-world elevation values.
  Required for heliacal visibility accuracy.

---

v1.9.0

The theme: classical muhurta sharpening — non-personal timing layer.
Sixteen computations from Muhurta Chintamani, Brihat Samhita,
Dharmasindhu, and standard panchangam authority — all surface as
additive properties on PanchangamDay and across the three per-day
MCP response paths (tool_get_panchangam, tool_get_muhurta,
tool_get_panchangam_range). No personal-astrology computations
(Dasha, Navamsa, Kuta match) — those are reserved for a separate
project. Engine math is additive only; the frozen core is untouched
and ICS subscriber feeds are byte-identical. +142 tests
(849 → 991 passed).

Added — Timing computations (16 features)

Foundational (Tasks 1–3):

• Ghati/vighati clock infrastructure — sunrise-anchored ghati
  scale (1 ghati = 1/60 ahoratri = sunrise→next-sunrise / 60).
  New GhatiClock and GhatiWindow dataclasses;
  PanchangamDay.ghati_clock exposes sunrise,
  next_sunrise, seconds_per_ghati. Foundation for all
  ghati-precision filters below. Module:
  telugu_panchangam/ghati.py.
• Moon's pada on the daily nakshatra span — PanchangamDay.nakshatra_pada
  (1..4), computed from each engine's own Moon longitude at sunrise.
  Drik value cross-pinned against gochara/positions.py:81 (canonical
  int(nak_pos * 4) % 4 + 1 formula).
• Ayanamsa as engine parameter — Drik now accepts
  ayanamsa: 'lahiri' | 'raman' | 'krishnamurti' | 'true_chitrapaksha'
  (default 'lahiri', byte-identical to prior behaviour).
  Surya Siddhanta and Vakya accept the parameter for API symmetry
  (their own mean-motion models drive sidereal positions; ayanamsa
  is a no-op there). New engines/utils.sidereal_longitude_with_ayanamsa
  helper restores Lahiri mode after each call to keep the hot-path
  @lru_cache consistent.

Ghati-precision filters (Tasks 4–6, ride on the ghati clock):

• Vishaghati windows — PanchangamDay.vishaghati: list[GhatiWindow].
  Per-nakshatra "poison ghatika" offsets from Muhurta Chintamani
  (27-entry table in karana_windows.VISHAGHATI_OFFSETS_GHATI);
  width = 4 vighatis (VISHAGHATI_WIDTH_VIGHATIS). Treated as
  inauspicious cuts in muhurta slot scoring, alongside Rahu Kalam etc.
  Exposed under inauspicious.vishaghati in MCP.
• Bhadra Mukha / Puchha — Vishti karana split into Mukha
  (first 5/16 of the half-tithi, hard-avoid) and Puchha
  (last 3/16, auspicious for contests/litigation per Muhurta
  Chintamani). PanchangamDay.bhadra_mukha /
  PanchangamDay.bhadra_puchha (GhatiWindow | None). Mukha is
  cut from slots; Puchha grants +2 to litigation slots that overlap it.
• Sankramana 16-ghati avoidance window —
  PanchangamDay.sankramana_avoidance: Window | None. 16 ghatis
  before + 16 ghatis after the Sun's exact rasi-ingress moment
  (~12h 48m total). Conservative 16+16 rule (vs 30+30 for
  Karkata/Makara cited in some sources) applied uniformly.
  Samskaras with skip_on_sankramana: True drop slots overlapping
  this window. New helper base._sankramanam_name_and_jd bisects
  for the exact ingress JD via find_crossing.

Nakshatra / yoga filters (Tasks 7, 13, 15):

• 5 Panchaka Nakshatras flag —
  PanchangamDay.in_panchaka_nakshatra: bool. True for Dhanishtha,
  Shatabhisha, Purva Bhadrapada, Uttara Bhadrapada, Revati. New
  activities cremation, construction_roof, wood_cutting have
  skip_on_panchaka_nakshatra: True. Distinct from Panchaka Rahita
  (modular-9 dosha) below.
• Anandadi 28 Yogas — PanchangamDay.anandadi_yoga: str | None.
  Vaaram × Moon's nakshatra → one of 28 named muhurta yogas
  (Ananda, Kalidanda, Dhumra, Dhata, … Vardhamana) per Muhurta
  Chintamani. Auspicious yogas grant +1 / inauspicious -1 in slot
  scoring; classification tables ANANDADI_AUSPICIOUS and
  ANANDADI_INAUSPICIOUS in special_yogas.py.
• Adho / Urdhva / Tiryan-Mukha nakshatra classification —
  PanchangamDay.nakshatra_mukha: str | None. Activity-conditioned
  bonus: Adho (construction_foundation, well_digging), Urdhva
  (coronation), Tiryan (travel). Table in
  nakshatra_filters.NAKSHATRA_MUKHA.

Solar / lunar maasa filters (Tasks 8, 11, 12):

• Khar-Maasa flag — PanchangamDay.is_khar_maasa: bool and
  PanchangamDay.khar_maasa_name: str | None ('Dhanur' when Sun
  in Dhanu rasi; 'Meena' when Sun in Meena). Samskara activities
  gain skip_on_khar_maasa: True. Note: codebase rasi spelling is
  'Dhanu' (not 'Dhanus').
• Adhika Maasa muhurta consumption — engine data was already
  in place since 1.0 (engines/base.py:121-143 names months as
  'Adhika <name>' / 'Nija <name>'), but find_muhurta ignored
  it. Samskaras now skip on day.maasam.startswith('Adhika ').
  Behaviour change: prior versions returned slots on Adhika
  days for wedding/upanayana/gruhapravesha/etc.; 1.9.0 returns
  [] with diagnostic 'Adhika Maasa — <activity> traditionally avoided'.
• Pitru Paksha 15-day window — PanchangamDay.is_pitru_paksha: bool.
  True for Bhadrapada Krishna paksha (15 days through Mahalaya
  Amavasya inclusive). Samskaras skip.

Graha-based filters — Drik only (Tasks 9, 10):

• Simha-Stha Guru / Shukra —
  PanchangamDay.simha_stha_guru: bool (hard skip on wedding
  via skip_on_simha_stha_guru) and
  PanchangamDay.simha_stha_shukra: bool (-2 score penalty via
  penalty_on_simha_stha_shukra). Strong south Indian custom
  (12-year Jupiter cycle). Surya Siddhanta and Vakya engines don't
  model outer planets — both fields stay False there.
• Guru / Shukra Maudhya (combustion / heliacal setting) —
  new MaudhyaInfo dataclass with graha, elongation_deg,
  combust, threshold_deg fields. Thresholds per Brihat Samhita:
  Jupiter 11°, Venus 10°. PanchangamDay.guru_maudhya and
  PanchangamDay.shukra_maudhya. wedding and upanayana gain
  skip_on_combust: ['Guru', 'Shukra'] — universal samskara rule
  across all regional traditions.

Activity-conditioned filter (Task 14):

• Disha Shoola — PanchangamDay.disha_shoola_direction: str | None
  (one of 'East' | 'West' | 'North' | 'South'). Classical weekday
  → blocked-travel-direction table. tool_find_muhurta gains a
  travel_direction parameter; when provided and matching the
  day's Disha Shoola, the travel activity drops slots.

Modular-9 dosha — the headline addition (Task 16):

• Panchaka Rahita — new PanchakaInfo dataclass with
  remainder (0..8), name ('Rahita' | 'Mrityu' | 'Agni' | 'Raja' | 'Chora' | 'Roga'), auspicious (bool), avoid_for
  (list of activity tags). PanchangamDay.panchaka_rahita populated
  at sunrise lagna. Slot-level recompute: muhurta slot scoring
  recomputes Panchaka using the slot's start-moment lagna (computed
  via personal/lagna_hora.get_lagna_transitions, cached once per
  day). Mrityu Panchaka caps slot tier (-3 to score, day-quality
  reason); other doshas apply -2 when the activity matches
  avoid_for. Module: telugu_panchangam/panchaka.py (sources:
  Muhurta Chintamani, Dharmasindhu).

Changed

MCP surface:

• All 19 new computation fields land on tool_get_panchangam,
  tool_get_muhurta, and tool_get_panchangam_range per-day dicts
  — consistent shape across all three tools (the
  multi-tool-serialization gap caught early in this round is now
  guarded by parity tests in each feature's test file).
• tool_get_panchangam, tool_get_panchangam_range,
  tool_get_graha_positions, tool_find_muhurta gain an optional
  ayanamsa: str = 'lahiri' parameter; the active ayanamsa is
  surfaced in each response.
• tool_find_muhurta gains an optional travel_direction: str | None
  parameter for Disha Shoola filtering on the travel activity.
• MCP server tool docstrings (mcp/server.py) updated to enumerate
  the new outputs.

Muhurta activity rules (personal/muhurta.ACTIVITY_RULES):

• 10 samskara activities (ceremony, wedding, engagement,
  naming, annaprasana, karnavedha, mundana, upanayana,
  vidyarambha, gruhapravesha) gained skip_on_sankramana,
  skip_on_khar_maasa, skip_on_adhika, skip_on_pitru_paksha
  rules — these days now correctly return no slots / explanatory
  diagnostic strings.
• 6 new activities added: litigation, cremation, coronation,
  construction_roof, wood_cutting, well_digging.
• wedding additionally gains skip_on_simha_stha_guru,
  penalty_on_simha_stha_shukra: -2, skip_on_combust: ['Guru', 'Shukra'].
• The activity-count test in tests/test_muhurta_finder.py is
  now test_all_30_activities_callable (was _all_24_…).

Behaviour-changing fixes caught during review:

• prefer_bhadra_puchha kwarg was extracted into ACTIVITY_RULES
  and declared on _evaluate_slot but never forwarded at the call
  site — the +2 Puchha bonus was a silent no-op. Now forwarded
  with a scoring-path regression test (a slot's reasons list
  must contain the bonus chip when overlap occurs).
• sidereal_longitude_with_ayanamsa set the global swisseph
  sid-mode but didn't restore it — restored on every call so the
  Lahiri @lru_cache hot-path never sees a stale mode.
• Daily pada formula initially used (moon_lon % nak_arc) / (nak_arc / 4.0) which mismatches the canonical gochara/positions.py
  form by 34 of 108 pada boundaries due to floating-point modulo
  instability — aligned to the canonical single-division form.
• One pre-existing test_vara_bonus_thursday_wedding was implicitly
  relying on a date where Jupiter happened to be 9.9° from the Sun
  (now correctly flagged combust); test rerouted to 2026-08-20.
• MCP serialization initially landed in only tool_get_panchangam
  for the first three features (Tasks 1, 2, 4); swept fix added
  every field to all three per-day MCP tool responses with parity
  tests baked into each subsequent task.

Notes

• Frozen-core compliance — no festival rules in engines/base.py
  touched, no generators/ics.py changes, no docs/ UI ...

v1.8.0 — Janma Lagna across the panchangam

v1.8.0 — Janma Lagna across the panchangam

Twelve PRs landing together (55, 58–68). The major additions are
around the natal ascendant (janma lagna): the engine now computes
lagna transitions and planetary horas; the muhurta finder scores
slots from both janma rashi and janma lagna independently; and
per-activity lagna-class preferences (Sthira / Chara / Dvisvabhava)
round out the classical Muhurta Chintamani guidance.

New features

• Lagna transitions + planetary horas as first-class engine output;
  exposed as get_daily_horas and get_lagna_transitions MCP tools.
• Per-city pre-computed lagna.json data layer + monthly cron.
• Hora strip + Lagna ribbon visible on the day card.
• Lagna kendra/trikona/Ashtama scoring in find_muhurta; Ashtama-
  Lagna joins Ashtama-Chandra as a tier-capping personal dosha.
• Optional janma_lagna per profile — strict Lagna Shuddhi runs
  alongside Chandra-Rashi mode (additive, both score).
• Per-activity lagna preferences: wedding -> Sthira, travel ->
  Chara, learning rites -> Dvisvabhava. Surfaces as a reason chip.
• Gochara verdicts from BOTH rashi and lagna lenses side by side.

Correctness fixes

• Amavasya and unrectified-Tara doshas correctly cap tier.
• tara_dosha tier cap restored in the JS UI (silent regression).
• "Parabhava Nama Samvatsara" devotee-preferred form everywhere.

Performance

• ~30% on Tarabalam multi-day searches via lru_cache on engine
  utils + a new calculate_bulk engine method.

Security

• DOM-XSS hardening at 7 user-name sinks in docs/index.html.

Infrastructure

• Pure muhurta scorer extracted into a Node-importable module with
  27 unit tests + CI parity job to prevent future Py/JS drift.

Engine surface change (owner sign-off recorded): minimal additive
touch in engines/utils.py (lru_cache on 4 pure swisseph wrappers)
and engines/base.py (default calculate_bulk that loops calculate).

PyPI: mcp-server-panchangam==1.8.0
Static site: panchangam.astrochaganti.com